Thread history
Viewing a history listing
| Time | User | Activity | Comment |
|---|---|---|---|
| 14:20, 25 March 2017 | Shervin.emami (talk | contribs) | New reply created |
(Reply to Image resolution in stereo_match test) |
| 11:31, 24 March 2017 | Kozuch (talk | contribs) | New reply created |
(Reply to Image resolution in stereo_match test) |
| 08:11, 15 April 2015 | Kozuch (talk | contribs) | New reply created |
(Reply to Image resolution in stereo_match test) |
| 16:26, 10 April 2015 | Shervin.emami (talk | contribs) | New reply created |
(Reply to Image resolution in stereo_match test) |
| 06:22, 10 April 2015 | Kozuch (talk | contribs) | New thread created |
Hi Kozuch, I was running the OpenCV sample code with sample images (ie: "opencv/samples/cpp/stereo_match.cpp" and "opencv/samples/gpu/stereo_match.cpp" on the images "left01.jpg" and "right01.jpg" in the cpp folder). So each of the 2 images would have been 640x480.
Hi again, I have one more question - did you test stereoBM with OpenCV4Tegra or with plain standard OpenCV? OpenCV4Tegra is said to have some tweaks for higher performance... I am actually interested in FPS figures for both OpenCV4Tegra and plain OpenCV. Thanks for your reply in advance!
Sorry I only tested one of them at the time, probably OpenCV4Tegra. It's true that OpenCV4Tegra includes some CPU and GPU optimizations, and it was updated many times since I first made these tests, so the performance difference might be different now than it used to. Also, the performance on Tegra X1 and X2 will probably be different than Tegra K1.
